Introduction
Glioblastoma multiforme (GBM) is one of the most malignant and aggressive tumors, and has a very poor prognosis and high recurrence rate with a mean survival time of less than 2 years even after the recent development of an intensive temozolomide (TMZ)-based treatment protocol (1,2). Thus, a novel therapeutic approach to controlling recurrence and overcoming resistance to treatment is crucially needed for glioblastoma patients. Novel therapeutic approaches to control tumor recurrence by targeting crucial signaling molecules such TGF-β (3), JAK-STAT (4–6) and WINT (7), adhesion molecules αβ integrin (8), and pro-angiogenic factors VEGF and VEGFR (9), have been tried. However, a new breakthrough has yet to be found.
Immunotherapy is also recognized as a therapeutic tool against human GBM, and preliminary studies trying to suppress recurrence have been reported (10–13).
Ardon et al (13) reported that dendritic cell (DC)-based vaccines against GBM had moderate effects in terms of a patient’s performance status (PS) or quality of life (QOL) in clinical phase I/II trial against advanced or newly diagnosed high-grade gliomas.
As to immunogenic antigens, several studies regarding glioma-specific cancer antigen including WT1, MAGE, gp100, HER2, which have been already used for clinical trial, were reported (14–16). WT-1 expression in glioblastomas was reported by Oji et al (14). Liu et al (16) reported that HER2, gp100 and MAGE-A1 mRNA expression was detected in 81.4, 46.5 and 39.5% of 43 GBM primary cell lines, and determined CTL epitope specific for each antigen. As to the CT antigen expression in gliomas, several studies have been reported (17–22). Debinski et al (17) showed the specific expression of IL-13RA2 gene, common CT antigen in high-grade gliomas. Recently, Syed et al (21) demonstrated that two CT antigens (MAGE-A1 and MAGE-A3) and gp100 may be candidate antigens in primary GBM cell lines using RT-PCR specific for 11 CT antigens and 4 melanocyte-differentiation antigens, however, those antigen expression was low-to-variable level. Furthermore, Sahin et al (22) analyzed expression of 7 CT antigens in 88 brain tumor specimen including meningiomas, and showed that SCP-1 exhibited the highest expression rate. However, a comprehensive expression analysis of CT antigens in high-grade gliomas is very limited.
In the present study, a comprehensive expression analysis of 54 CT antigens and 13 glioma-associated antigens were performed using a quantitative PCR in 17 high-grade glioma-derived primary cell lines, and the co-relation of antigen expression to overall survival and O6-methylguanine-DNA-methyltransferase (MGMT) expression was also investigated.
Materials and methods
Establishment of primary GB cell lines from high-grade glioma patients
Glioma tumor samples were obtained from surgical resections by neurosurgeons in Shizuoka Cancer Center. The clinical research using tumor tissues from glioma patients was approved by the Institutional Review Board of Shizuoka Cancer Center, Shizuoka, Japan. All patients gave written informed consent. Seventeen high-grade glioma patients, who are sources of established primary glioma cell lines, are characterized in Table I.
Tumors were dissociated by teasing with forceps to make a single cell suspension, and plated in 25 cm2 culture flask in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen), penicillin, streptomycin and gentamycin (Invitrogen) for serum-derived GB cell line. Glioblastoma cell lines (A172, T98G, LN-18, U87, U118, U138 and U373) was purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA) and maintained in DMEM containing 10% FBS plus penicillin and streptomycin.
Antibodies and reagents
Antibodies against 6 CT antigens (MAGE-A12, BAGE, DDX43, IL-13RA2, CTAGE1 and CASC5) were purchased from the following sources; Abnova for MAGE-A12, LSBio Inc. for BAGE, Proteintech for DDX43 and IL-13RA2, Santa Cruz Biotechnology for CTAGE1, Bethyl Laboratories Inc. for CASC5. Antibodies against 10 glioma-associated antigens (HER2, Ki-67, Podplanin, gp100, GFAP, MGMT, Nestin, Olig2, S100 and CD133) were purchased from the following sources; Dako for HER2, Ki-67 and gp100, Santa Cruz Biotechology for Podplanin, AbD Serotec for GFAP, Abcam for MGMT and IBL for Nestin, Olig2 and CD133.
Quantitative polymerase chain reaction (qPCR) analysis
The real-time PCR analysis of CT antigen and glioma-associated genes using the 7500 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) was performed as previously described. Briefly, 54 CT antigen-specific PCR primers (from CT1.1 to CT50; shown in Table II) or 13 glioma-associated gene-specific PCR primers (HER2, Ki-67, Podplanin, gp100, GFAP, MGMT, Nestin, Olig2, S100, CD133, WT1, MART-1 and Tyrosinase) and TaqMan probes were purchased from Applied Biosystems. Total RNAs were extracted from each glioma cell line or primary cells derived from high-grade gliomas. Complementary DNA was synthesized from 100 ng of the total RNA and quantitative PCRs were carried using a TaqMan RNA-to-CT 1-Step kit (Applied Biosystems). As a negative control for tumor RNA, total RNA from the normal fetal brain (platinum total RNA) was purchased from Takara Bio Inc. (Otsu, Shiga, Japan).
Immunohistochemistry (IHC)
Two representative cases, GB-case no. 16 and 9-derived resected tumors were fixed with formalin solution and used for IHC analysis of CT antigens and glioma-associated antigens, respectively. Hematoxylin-eosin staining was performed according to the manufacturer’s instructions. For immunostaining, antibodies against 6 CT antigens (MAGE-A12, BAGE, DDX43, IL-13RA2, CTAGE1 and CASC5) and antibodies against 10 glioma-associated antigens (HER2, Ki-67, Podplanin, gp100, GFAP, MGMT, Nestin, Olig2, S100 and CD133) were used as primary antibody. Goat anti-mouse IgG antibody or goat anti-rabbit IgG was used as secondary antibody. Horseradish peroxidase (HRP) and hydrogen peroxide were utilized for making color according to the manufacturer’s instructions (Vectastatin ABC kit; Vector Laboratories, Burlingame, CA, USA).
Statistical analysis
The overall survival analysis based on the REMBRANDT dataset regarding target antigen genes was examined by comparing differences in median survival time (MST) via the Kaplan-Meier method. A comparative analysis of survival times between groups was then performed using the log rank test. Values of P<0.05 were considered statistically significant. For the correlation analysis to MGMT expression, the correlation coefficient, r, was calculated and statistical difference was analyzed using the Pearson’s correlation test. Values of P<0.05 were considered statistically significant.
Results
Establishment of high-grade glioma primary cell lines
Seventeen serum-derived primary cell lines were established. These cell lines consisted of 3 anaplastic oligodendrogliomas, 1 anaplastic astrocytoma and 13 GBMs. Histologically, 13 of the 17 cases were WHO grade IV. Morphological appearances seem to be mainly adherent, but different among 17 primary cell lines ranging from round to fibroblast-like shape. Heterogeneous cell populations were seen, where adherent and floating cells were mixed (Fig. 1).
Quantitative PCR analysis of CT antigen or glioma-associated gene mRNA in glioma primary cell lines
A quantitative PCR using 54 cancer-testis (CT) antigen-specific primers showed that 36 CT antigens were positive in at least 1 of 17 serum-derived cell lines, and 17 antigens were positive in >50% cell lines (Table III). Impressively, 6 genes (BAGE, MAGE-A12, CSAC5, CTAGE1, DDX43 and IL-13RA2) were detected in all the cell lines. On the other hand, the expression of other 13 glioma-associated antigens than CT genes were also investigated, and 10 genes were detected in >70% of the cell lines (Table IV). Glioma-specific antigens CD133, S100, Olig2, Podplanin and Nestin were detected in most primary cells as well as commercially available cell lines. Notably, gp100 and HER2, common cancer antigens, were also positively expressed in all cell lines as well as the regular ones. MGMT was positive in all cell lines except one.
Immunohistochemistry
The expression of CT antigen and glioma-associated antigen genes with a high frequency were verified in IHC analysis using GB-case 16-derived resected tumor specimen. In 6 CT antigens (MAGE-A12, BAGE, DDX43, IL-13RA2, CTAGE1 and CASC5), all antigen proteins other than CTAGE1 were positively expressed in the tumor from patient (Fig. 2). On the other hand, 10 glioma-associated antigens (HER2, Ki-67, Podplanin, gp100, GFAP, MGMT, Nestin, Olig2, S100 and CD133) staining showed that all antigen proteins besides MGMT, CD133, HER2 and gp100 were detected in the tumor from GB-case 9 (Fig. 3).
CT antigen gene expression compared with normal brain tissue
Gene expression levels of 6 CT antigens in primary high-grade glioma cell lines were significantly upregulated in CASC5 and CTAGE1 compared with normal brain (Fig. 4). In particular, CASC5 expression level was higher almost 10-fold than normal brain. Unexpectedly, IL-13RA2 mean expression level was lower than normal level.
Correlation of CT antigen expression to overall survival and MGMT expression
The correlation of antigen expression level to overall survival time was analyzed using REMBRANDT database of the National Cancer Institute in the 17 CT antigens and 10 glioma-associated genes which were highly expressed in glioma primary cell lines. The expressions of IL-13RA2, HER2, Ki-67, podplanin, Olig2 and CD133 were inversely correlated to overall survival time in high-grade glioma patients given a standard chemo-radiation after surgery (P<0.01; Table V). Another correlation analysis of CT antigen and glioma-associated gene to MGMT expression demonstrated that 4 genes including DDX43, TDRD1, HER2 and gp100 had significant positive relationship to MGMT expression. Particularly, HER2 and gp100 genes showed very high coefficient (r=0.751 and 0.894, respectively; Table VI).
Discussion
Glioblastoma multiforme (GBM) is one of the most malignant and aggressive tumors, and even with an intensive treatment including surgery, radiation and chemotherapy, the median survival time is about 1 year and few cases can survive more than 2 fears (1,2). Therefore, novel therapeutic approach to control the recurrence and overcome the resistance to treatment is crucially needed in glioblastoma patients.
A prerequisite for the successful immunotherapy is the identification of suitable tumor antigens specifically expressed in cancer tissues and its clinical application as an advanced translational research. Renkvist et al (23) classified human tumor antigens recognized by T cells into 6 categories; i) HLA class I-restricted cancer/testis antigens, ii) differentiation antigens, iii) widely expressed antigens, iv) tumor-specific antigens, v) HLA class II-restricted antigens and (vi) fusion proteins. Particularly, CT antigen and melanoma-associated differentiation antigen have been focused on, and several peptides including MAGE family, NY-ESO, IL-13RA2, MART-1, gp100 and tyrosinase were effectively used in clinical vaccinations for advanced melanoma and glioma patients (24–28).
To develop novel glioma-specific antigens, in the present study, we performed a comprehensive expression analysis of CT antigen and glioma-associated genes in patients-derived primary glioma cell lines using a quantitative PCR. A quantitative PCR using 54 cancer-testis (CT) antigen-specific primers showed that 36 CT antigens were positive in at least 1 of 17 serum-derived cell lines, and 17 antigens were positive in >50% cell lines. Particularly, 6 genes (BAGE, MAGE-A12, CSAC5, CTAGE1, DDX43 and IL-13RA2) were detected in all cell lines. Sahin et al (22) reported the frequency of selected 7 CT antigen gene expression in 88 human brain tumor specimen using RT-PCR. The highest frequency was 40% in SCP-1 gene expression, and 23 astrocytomas expressed at least one CT gene. Another RT-PCR study by Syed et al (21) using 45 glioma-frozen tissues revealed that the highest CT antigen expression rates were MAGEA-3 (22%) and MAGEA-1 (16%) in 11 CT genes.
Our highest expression rate was 100% in 6 CT genes, and all primary cell lines showed expression in at least 14 CT antigen genes, thus showing significantly better results compared to the previous study. Possible reasons why our expression frequency was higher is speculated as; i) higher purity of tumor cells because newly established serum-derived primary cell cultures were used for analysis, ii) the number of CT genes analyzed was higher, iii) validated quantitative PCR primer. On the other hand, the expression of other 13 glioma-associated antigens than CT genes were also investigated, and 10 genes were detected in more than 70% of the cell lines. Specifically, glioma-assocated genes such as HER2, podoplanin and gp100 were positive in all the cell lines.
HER2 and gp100 as melanocyte-differentiation antigen (MDA) were reported to be highly expressed in malignant melanoma and glioma tissues as well as WT1. Interestingly, because both melanocytes and glial cells are derived from the neuroectoderm, it is not surprising that they express common MDA and CT antigen genes (29).
In the present study, correlation analysis of CT antigen and glioma-associated genes to overall survival time and MGMT expression was performed. IL-13RA2 and HER2 genes were demonstrated to be prognostic factor candidates for a poor survival (Table V), which may suggest that prognostic analysis can contribute to the understanding the novel function of CT antigen genes. Both target genes are well known to be expressed in GBM tumors, therefore specific clinical immunotherapy against these molecules has been performed in a small size of trail design. Specifically, obvious evidence that the downregulation of IL-13RA2 and HER2 antigen expression by therapeutic agents may contribute to a good prognosis has yet to be obtained. Furthermore, few studies regarding the correlation between CT antigens and survival time have been reported to date. That is why our observations should be stressed in terms of novelty.
Additionally, we analyzed the correlation of CT antigen genes to MGMT expression. Interestingly, 4 antigen gene candidates positively correlated to MGMT expression, were identified. In particular, HER2 and gp100 showed a high coefficient (r=0.751 and 0.894, respectively). MGMT expression is considered to be regulated by DNA methylation mechanism, and crucial in terms of determining the chemosensitivity of high-grade gliomas to temozolomide (30). It is well-known that high expression of MGMT is closely associated with poor prognosis in high-grade gliomas, and DNA-methylation of MGMT can be a possible approach to MGMT reduction (31,32).
If some CT antigens are linked to MGMT gene in DNA-demethylation to promoter area, epigenetic approach to DNA methylation of CT genes could be a novel therapy, resulting in a therapeutic effect on high-grade gliomas. However, there is no evidence for correlation of both gene promoters, and clarification of the question will need more investigation. Natsume et al (33) reported that the DNA-demethylating agent 5-aza-2′-deoxycytidine (5-aza-CdR) remarkably reactivated NY-ESO-1 gene expression and demonstrated that 5-aza-CdR-induced expression of epigenetically silenced CT antigen gene might be a new strategy for immunotherapy against high-grade glioma.
Little is known about the precise function of CT antigen genes. Notably, Low et al (34) reported a novel approach to evaluating the function of CT gene using knockdown technology against neural stem cell-derived CT antigen, and showed some CT antigen genes were associated with the stem cell differentiation.
Finally, in the present study, we analyzed expression of 54 CT antigens using a quantitative PCR. In near future, CT antigen repertoire will be expanded up to 100, and more efficient biomarkers and prognostic markers specific for high-grade glioma should be collected, where effective target antigens can be determined for developing novel immunotherapy and antibody medicines.